Detecting dispersal: A spatial dynamic occupancy model to reliably quantify connectivity across heterogeneous conservation landscapes

Abstract

Connectivity shapes species distribution, spatial population dynamics and genetic structure, and is critical for conservation. It is imperative to reliably identify factors that limit connectivity across heterogeneous, fragmented landscapes. Viewing connectivity as the persistence of movement in space from source to destination—along potential corridors simulated from correlated random walks—we adapt the dynamic occupancy modeling framework to present an approach to reliably quantify connectivity. This approach allowed us to estimate the probability of spatial persistence of movement along potential corridors, while accounting for imperfect detection of animal use of the matrix. We use simulations to test our model, and data from sign surveys of Asian elephant Elephas maximus space use in Garo Hills, Northeast India, as a practical application of the method. Distance to forests and ruggedness predominantly shaped elephant connectivity patterns. Negative effects of distance to forests were intensified at the onset of dispersal, and in the more disturbed part of our landscape. We mapped locations critical to maintaining connectivity in our study landscape. We demonstrate that the widely used occupancy modeling approach, when combined with appropriate field data collection, permits explicit assessment of matrix impacts on connectivity while accounting for imperfect detection of animal movement. In so doing, we highlight the value of the approach in enabling inference on where animals move, as well as why. Obtaining a reliable understanding of factors shaping connectivity is central to understanding and predicting species responses to environmental change, thereby facilitating effective long-term conservation in fragmented landscapes.